Hello, students. Today, I'm going to use clusterEnG to demonstrate the usage of two publishing based algorithms, K- means and K-medoids. As we have learned from the class, K-Means is a very popular publishing based clustering algorithm. To partition the given data points into k clusters. K-means first selects k clusters centers randomly. Then, for every data point K-means computes distances to the current k centers. Under sizes to the closest center. After that, K-means needs to update the K centers by computing the mean of each cluster, such a process is repeated until the K-centers do not change. The K-matter is algorithm is very similar to the K-means algorithm. With the only difference that we are not updating the center of each cluster we do not compute the mean of all the current data points instead, we choose one data point from the cluster that minimize the with-in cluster variance. On clustering remember that there are three K steps. Data preparation, Algorithm choosing and Result visualization. In this two hierarchical, lastly, used simple data set that consist of 15 to data points, after uploading the data which use K-means as a cluster algorithm. Also, a very important step is choosing a number of clusters for K-means. Here, we say this parameter to straight. But the value of this parameter depends on the specific specification and the data set used in practice, we can try out different parameter values and determine the triple one. The running k means with k equal three we cluster the data points into three clusters. As we can see from these plots the original 15 data points are separated into three very clear clusters. We can also try out different k values. Let's say if we said k equals five, and k means again on the same set of sides. Then, we're actually imposing finer regularity on the classroom process. As we can see from the new plot. The data points that our ration may belong to the same class are now separated into two different clusters. The k-medoids algorithm is very similar to k-means. You can still use the process to run k-medoids on the same div sets. And you'll find out that the clustering results will be very similar. You can get more information about those two algorisms. You can also visit a tutorial of that page. So that concludes today's tutorial. Thanks for watching. [MUSIC] [MUSIC]